Publications of the Astronomical Society of the Pacific, 130:114401 (14pp), 2018 November https://doi.org/10.1088/1538-3873/aadf6f © 2018. The Astronomical Society of the Pacific. All rights reserved. Printed in the U.S.A.

A Framework for Prioritizing the TESS Planetary Candidates Most Amenable to Atmospheric Characterization Eliza M.-R. Kempton1,2 , Jacob L. Bean3 , Dana R. Louie1, Drake Deming1, Daniel D. B. Koll4, Megan Mansfield5, Jessie L. Christiansen6, Mercedes López-Morales7, Mark R. Swain8, Robert T. Zellem8, Sarah Ballard9,45, Thomas Barclay10,11, Joanna K. Barstow12, Natasha E. Batalha13, Thomas G. Beatty14,15, Zach Berta-Thompson16, Jayne Birkby17, Lars A. Buchhave18, David Charbonneau7, Nicolas B. Cowan19,20, Ian Crossfield9, Miguel de Val-Borro21,22, René Doyon23, Diana Dragomir9,46, Eric Gaidos24, Kevin Heng25, Renyu Hu8, Stephen R. Kane26, Laura Kreidberg7,27, Matthias Mallonn28, Caroline V. Morley29, Norio Narita30,31,32,33,34, Valerio Nascimbeni35, Enric Pallé34,36, Elisa V. Quintana10, Emily Rauscher37, Sara Seager38,39, Evgenya L. Shkolnik40, David K. Sing41, Alessandro Sozzetti42, Keivan G. Stassun43, Jeff A. Valenti13, and Carolina von Essen44 1 Department of , University of Maryland, College Park, MD 20742, USA; [email protected] 2 Department of Physics, Grinnell College, 1116 8th Avenue, Grinnell, IA 50112, USA 3 Department of Astronomy & Astrophysics, University of Chicago, 5640 S. Ellis Avenue, Chicago, IL 60637, USA 4 Department of , Atmospheric, and Planetary Sciences, Massachusetts Institute of Technology, Cambridge, MA 02139, USA 5 Department of Geophysical Sciences, University of Chicago, 5734 S. Ellis Avenue, Chicago, IL 60637, USA 6 Caltech/IPAC-NASA Science Institute, MC 100-22 Pasadena CA 91125, USA 7 Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 01238, USA 8 Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Drive, Pasadena, CA 91109, USA 9 Kavli Institute for Astrophysics and Space Research, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139, USA 10 NASA Goddard Space Flight Center, 8800 Greenbelt Road, Greenbelt, MD 20771, USA 11 University of Maryland, Baltimore County, 1000 Hilltop Cir, Baltimore, MD 21250, USA 12 Department of Physics and Astronomy, University College London, London, UK 13 Space Telescope Science Institute, 3700 San Martin Dr., Baltimore, MD 21218, USA 14 Department of Astronomy & Astrophysics, The Pennsylvania State University, 525 Davey Lab, University Park, PA 16802, USA 15 Center for and Habitable Worlds, The Pennsylvania State University, 525 Davey Lab, University Park, PA 16802, USA 16 Department of Astrophysical and Planetary Sciences, University of Colorado, Boulder, CO 80309, USA 17 Anton Pannekoek Institute for Astronomy, University of Amsterdam, Science Park 904, 1098 XH Amsterdam, The Netherlands 18 DTU Space, National Space Institute, Technical University of Denmark, Elektrovej 328, DK-2800 Kgs. Lyngby, Denmark 19 Department of Physics, McGill University, 3600 rue University, Montréal, QC, H3A 2T8, Canada 20 Department of Earth & Planetary Sciences, McGill University, 3450 rue University, Montreal, QC, H3A 0E8, Canada 21 NASA Goddard Space Flight Center, Astrochemistry Laboratory, 8800 Greenbelt Road, Greenbelt, MD 20771, USA 22 Department of Physics, Catholic University of America, Washington, DC 20064, USA 23 Institut de Recherche sur les Exoplanètes, Départment de Physique, Université de Montréal, Montréal, QC H3C 3J7, Canada 24 Department of Geology & Geophysics, University of Hawai‘i at Mänoa, Honolulu, HI 96822, USA 25 University of Bern, Center for Space and Habitability, Gesellschaftsstrasse 6, CH-3012, Bern, Switzerland 26 Department of Earth Sciences, University of California, Riverside, CA 92521, USA 27 Harvard Society of Fellows 78 Mt. Auburn Street, Cambridge, MA 02138, USA 28 Leibniz-Institut für Astrophysik Potsdam, An der Sternwarte 16, D-14482 Potsdam, Germany 29 Department of Astronomy, Harvard University, 60 Garden St, Cambridge MA 02138, USA 30 Department of Astronomy, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan 31 JST, PRESTO, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan 32 Center, NINS, 2-21-1 Osawa, Mitaka, Tokyo 181-8588, Japan 33 National Astronomical Observatory of Japan, NINS, 2-21-1 Osawa, Mitaka, Tokyo 181-8588, Japan 34 Instituto de Astrofísica de Canarias (IAC), E-38205 La Laguna, Tenerife, Spain 35 Dipartimento di Fisica e Astronomia, “G. Galilei”, Università degli Studi di Padova, Vicolo dell’Osservatorio 3, I-35122 Padova, Italy 36 Departamento de Astrofísica, Universidad de La Laguna (ULL), E-38206, La Laguna, Tenerife, Spain 37 Department of Astronomy, University of Michigan, 1085 S. University Avenue, Ann Arbor, MI 48109, USA 38 Massachusetts Institute of Technology, Department of Earth, Atmospheric, and Planetary Sciences, 77 Massachusetts Avenue, Cambridge, MA 02139, USA 39 Massachusetts Institute of Technology, Department of Physics, 77 Massachusetts Avenue, Cambridge, MA 02139, USA 40 School of Earth and Space Exploration, Arizona State University, Tempe, AZ, 85287, USA 41 Astrophysics Group, Physics Building, Stocker Road, University of Exeter, Devon EX4 4QL, UK 42 INAF—Osservatorio Astrofisico di Torino, Via Osservatorio 20, I-10025 Pino Torinese, Italy 43 Vanderbilt University, Department of Physics & Astronomy, 6301 Stevenson Center Lane, Nashville, TN 37235, USA 44 Stellar Astrophysics Centre, Department of Physics and Astronomy, Aarhus University, Ny Munkegade 120, DK-8000 Aarhus C, Denmark Received 2018 May 6; accepted 2018 September 6; published 2018 September 27

45 MIT Torres Fellow for Exoplanetary Science. 46 NASA Hubble Fellow.

1 Publications of the Astronomical Society of the Pacific, 130:114401 (14pp), 2018 November Kempton et al.

Abstract A key legacy of the recently launched the Transiting Exoplanet Survey Satellite (TESS) mission will be to provide the astronomical community with many of the best transiting exoplanet targets for atmospheric characterization. However, time is of the essence to take full advantage of this opportunity. The James Webb Space Telescope (JWST),although delayed, will still complete its nominal five year mission on a timeline that motivates rapid identification, confirmation, and mass measurement of the top atmospheric characterization targets from TESS.BeyondJWST, future dedicated missions for atmospheric studies such as the Atmospheric Remote-sensing Infrared Exoplanet Large-survey (ARIEL) require the discovery and confirmation of several hundred additional sub-Jovian size (Rp<10 R⊕) orbiting bright stars, beyond those known today, to ensure a successful statistical census of exoplanet atmospheres. Ground- based extremely large telescopes (ELTs) will also contribute to surveying the atmospheres of the transiting planets discovered by TESS. Here we present a set of two straightforward analytic metrics, quantifying the expected signal-to- noise in transmission and thermal emission spectroscopy for a given , that will allow the top atmospheric characterization targets to be readily identified among the TESS planet candidates. Targets that meet our proposed threshold values for these metrics would be encouraged for rapid follow-up and confirmation via radial velocity mass measurements. Based on the catalog of simulated TESS detections by Sullivan et al., we determine appropriate cutoff values of the metrics, such that the TESS mission will ultimately yield a sample of ∼300 high-quality atmospheric characterization targets across a range of planet size bins, extending down to Earth-size, potentially habitable worlds. Key words: planets and satellites: atmospheres – planets and satellites: detection Online material: color figures

1. Introduction Figure 1 shows the known transiting exoplanets that are favorable targets for atmospheric characterization based on The Transiting Exoplanet Survey Satellite (TESS) is poised their expected S/N for transmission spectroscopy.51 While to revolutionize the exoplanet field by completing the census of there are a substantial number of giants that are good targets, close-in transiting planets orbiting the nearest stars (Ricker there are very few known planets smaller than 10 R⊕ that are et al. 2015). The planets discovered by TESS will be among the suitable for this type of atmospheric study. The atmospheric best targets for atmospheric characterization, owing to the large characterization community has the ambition to use JWST and signal-to-noise (S/N) obtainable based on the brightness and the ELTs to characterize many tens of planets, including a push relatively small sizes of their host stars. Most notably, TESS is toward temperate and rocky worlds (Snellen et al. 2013; Rodler designed to detect many hundreds of sub-Jovian size planets & López-Morales 2014; Cowan et al. 2015). Furthermore, that are substantially better atmospheric characterization targets hundreds of planets over a wide range of masses and radii will than those detected by the Kepler satellite (e.g., Thompson ultimately be needed for future dedicated exoplanet atmosphere et al. 2018). The atmospheric characterization studies enabled missions like the recently selected Atmospheric Remote-sensing by the TESS mission will round out our understanding of Infrared Exoplanet Large-survey (ARIEL) concept (Tinetti exoplanet atmospheres in the Neptune- down to Earth-size et al. 2016; see Section 2 for more discussion). regime and may even extend to habitable worlds. While TESS is positioned to deliver the planets needed for The backdrop for the TESS mission is the dramatic increase atmospheric characterization efforts, substantial follow-up in our capability to probe the atmospheres of transiting work is needed to turn its detected planet candidates into bona exoplanets that is expected over the coming decade. The fide planets that are appropriate targets for atmospheric launch of the James Webb Space Telescope (JWST) is highly observations. Following TESS detection, the essential follow- anticipated, and construction of the next generation of up steps for this process include improved characterization of extremely large telescopes (ELTs) on the ground is already the host star (e.g., Stassun et al. 2018), additional transit underway. A primary science driver for both JWST and the observations to increase the precision on the orbital ephemer- ELTs is the characterization of exoplanet atmospheres,47,48,49,50 ides (e.g., Kane et al. 2009), validation or confirmation of and the planets discovered by the TESS mission will be vital to planetary nature, and planet mass measurement. realizing this vision. Typically the most resource-intensive component of candidate ( ) fi 47 https://jwst.nasa.gov/science.html follow up is the radial velocity RV con rmation and 48 http://www.gmto.org/Resources/GMT-SCI-REF-00482_2_GMT_ measurement of planet mass. While planet validation techniques Science_Book.pdf 49 https://www.eso.org/sci/facilities/eelt/science/doc/eelt_sciencecase.pdf 51 Throughout this paper, we use the properties of known transiting exoplanets 50 https://www.tmt.org/page/exoplanets and their host stars from TEPCat: http://www.astro.keele.ac.uk/jkt/tepcat/.

2 Publications of the Astronomical Society of the Pacific, 130:114401 (14pp), 2018 November Kempton et al.

of planets orbiting M-dwarf hosts, Barclay et al. (2018), who recalculated the planetary yields using the actual TESS Input Catalog (TIC) of target stars, and Huang et al. (2018), who also used the TIC stars but with updated parameters from the Gaia Data Release 2 along with improved treatment of multi-planet systems and TESS noise systematics. The more recent results are mostly in line with the findings of Sullivan et al. (2015), with a couple of differences that have implications for our current work. These differences are (1) The M- occurrence rates found by Ballard (2018) and Huang et al. (2018) are higher than those previously reported by up to 50%, and (2) Bouma et al. (2017) and Barclay et al. (2018) both report an overestimation of the number of and super- Earths in the Sullivan et al. (2015) work that resulted from an error in the latter’s calculations. Additionally, none of these works account for the paucity of planets with Rp≈1.5 R⊕ associated with the planetary radius gap identified by Fulton ( ) ( ) Figure 1. Masses and radii of known transiting exoplanets that are good targets et al. 2017 and Van Eylen et al. 2018 , which could lead to a for atmospheric characterization. Formally, these are selected as the known small over-prediction of such planets in each of the aforemen- planets with estimated transmission metric values greater than 50, as defined in tioned catalogs. Section 3. In this work, we employ the Sullivan et al. (2015) catalog, (A color version of this figure is available in the online journal.) while noting the discrepancies with more recent simulated TESS yield calculations. As a check for the effect of these differences, we have repeated the key steps of our analysis exist that bypass this step and do not require RV mass using the Barclay et al. (2018) catalog, and we discuss those ( ) measurements Torres et al. 2011 , precise mass determinations outcomes in Section 5. Using the Sullivan et al. (2015) results have been shown to be fundamental to correctly interpreting for our primary analysis allows us to build on the Louie et al. atmospheric observations of exoplanet transmission spectra (2018) simulations of JWST/Near Infrared Imager and Slitless ( ) Batalha et al. 2017a . Furthermore, well-constrained radial Spectrograph (NIRISS) transit observations of the planets in velocity orbits are needed to predict secondary eclipse times. that catalog. We identify cutoffs to select the top atmospheric Mass measurements are a key component of the TESS Level-1 characterization targets based on the expected S/N of the mission requirements, which specify that 50 planets with simulated TESS planets in transmission and emission spectrosc- < Rp 4 R⊕ have their masses measured through RV followup opy. Our methodology and threshold criteria for identifying the ( ) Ricker et al. 2015 . The recent delays of the JWST launch have best atmospheric characterization targets from TESS are afforded the exoplanet community with an additional time described below. We concentrate mainly on JWST observa- cushion for candidate follow-up efforts. Nevertheless, the bility, with the expectation that the results would be community still must act quickly to ensure that the best qualitatively similar for calculations done specifically for atmospheric characterization targets are confirmed and weighed ground-based telescopes or ARIEL. on the timeline of the prime JWST mission. Our goal here is to motivate a set of threshold criteria to identify the TESS planet candidates that are expected to be most 2. Sample Selection amenable to atmospheric characterization, and therefore merit We consider three samples of planets for our analysis rapid RV follow up. We base our selection thresholds on our of atmospheric observability, two in terms of their observability understanding of the expected mission planet yields from the for transmission spectroscopy and one in terms of its simulated TESS catalog of Sullivan et al. (2015). The Sullivan observability for emission spectroscopy. The two transmission et al. (2015) catalog is one realization of the TESS planet spectroscopy samples are (1) a large sample of planets across a detection outcomes based on published occurrence rate range of planet sizes and (2) a sample of small planets in and statistics (Fressin et al. 2013; Dressing & Charbonneau 2015) near the habitable zones of their host stars. The emission and a galactic stellar population model (Girardi et al. 2005). spectroscopy sample is composed of planets that have sizes The TESS exoplanet yields have been re-examined more consistent with a terrestrial composition. The properties of recently by a number of authors including Bouma et al. (2017), these samples are described in more detail in the following who investigated strategies for an extension of the TESS subsections, and then appropriate threshold criteria to deliver mission, Ballard (2018), who studied yields and multiplicities these samples are identified in Section 4.

3 Publications of the Astronomical Society of the Pacific, 130:114401 (14pp), 2018 November Kempton et al.

2.1. Statistical Sample Based on simulations performed for the Fast Infrared Exoplanet Spectroscopy Survey Explorer (FINESSE) mission proposal (Bean & FINESSE Science Team 2017), we postulate that one goal of atmospheric characterization efforts over the next decade should be a transmission spectroscopy survey of approximately 500 planets so that statistical trends can be revealed. These simulations indicate that on order of 500 planets are needed to accurately discern population-wide properties like the relationship between atmospheric metallicity and planetary mass and differences between stellar and planetary atmospheric abundance ratios (e.g., C/O) in the face of the diversity predicted by planet formation models (Fortney et al. 2013; Mordasini et al. 2016). We focus first on transmission spectroscopy because this is expected to be the prime mode for exoplanet atmospheric observations and provides the best sensitivity to a wide range of planets. To create a 500 planet statistical sample we will need Figure 2. Histogram of the radii for the known transiting exoplanets that are roughly 300 new planets from TESS that sample the radius good targets for atmospheric characterization with transmission spectroscopy ( ) TESS parameter space Rp<10 R⊕. We therefore take the catalog of orange with the potential best planets from our statistical sample ( ) simulated TESS detections from Sullivan et al. (2015) and green . ( fi ) divide it into the following four planet-size bins: A color version of this gure is available in the online journal.

1. Terrestrials: Rp<1.5 R⊕, 2. Small sub-Neptunes: 1.5<Rp<2.75 R⊕, scalable to the achievable S/N for ground-based observations, 3. Large sub-Neptunes: 2.75<Rp<4.0 R⊕, and (see more discussion in Section 4.1). 4. Sub-Jovians: 4.0<Rp<10 R⊕. To build a sample of ∼300 total targets, from the large and small sub-Neptune bins (total number N=578 and N=1063, We aim to select the best simulated planets from each of the respectively), we select the top 100 planets each. There are only four planet size bins to be considered for RV follow up and 100 (exactly) planets in the sub-Jovian bin, and we ultimately eventual atmospheric characterization (the latter assuming the recommend to follow up the best 50 (see Section 4.1). From the RV mass and stellar activity metrics render the candidate a bin, in which the total number of expected high-quality target for transmission spectroscopy). We initially TESS discoveries drops off because the transit depths approach identify high-quality atmospheric characterization targets by the mission’s detection threshold, we select the top quintile using the predicted transmission spectroscopy S/N according planets (37 out of a total N=192). The combined sample of to the results of Louie et al. (2018) from their end-to-end 287 planets with R <10 R⊕ from the four size bins constitutes JWST/NIRISS simulator for the case of a 10 hr observing p “ ” campaign. The Louie et al. (2018) S/N values are calculated our statistical transmission spectroscopy sample. The histo- ( for the detection of spectral features (i.e., difference from a flat gram of the planet radii for the known planets the same ones ) line) integrated over the entire NIRISS bandpass, which is shown in Figure 1 and this statistical sample is shown in 0.6–2.8 μm for most targets and 0.8–2.8 μm for bright targets Figure 2. that require the use of a smaller subarray. We concentrate on NIRISS because a precise simulation of 2.2. Small Temperate Sample the Sullivan et al. (2015) catalog has already been done for this instrument, and because NIRISS gives more transmission In addition to the statistical sample of the most easily spectroscopy information per unit of observing time compared characterizable planets of a given size, we also consider a to the other JWST instruments for a wide range of planets and sample of planets in and near the liquid water habitable zone as host stars (Batalha & Line 2017;Howeetal.2017).However,it targets for transmission spectroscopy measurements. Following is worth keeping in mind that other JWST instruments may be Sullivan et al. (2015), we delineate this sample as being planets more suitable for observations in certain corners of parameter with insolation values of Sp=0.2–2.0 times the Earth’s space. For example, Morley et al. (2017) suggested the use of insolation (S⊕) and Rp<2.0 R⊕. A total of 60 simulated NIRSpec for observations of small planets orbiting cool, nearby planets from the Sullivan et al. (2015) catalog meet the criteria stars. Also, the predicted NIRISS S/N should be approximately of this “small temperate” transmission spectroscopy sample,

4 Publications of the Astronomical Society of the Pacific, 130:114401 (14pp), 2018 November Kempton et al. and we perform a further down-selection of this sample based 3.1. Transmission Metric on transmission spectroscopy detectability (see Section 4.2). For each planet in the statistical and small temperate We note that the lower insolation boundary (0.2 S⊕) of the samples, we calculate a transmission spectroscopy metric small temperate sample is commensurate with the outer edge of (TSM) that is proportional to the expected transmission the habitable zone for low-mass stars as calculated by spectroscopy S/N, based on the strength of spectral features Kopparapu et al. (2013), while the higher insolation boundary 2 (µRHp R , where H is the atmospheric scale height) and the of this sample (2 S⊕) is well interior to the inner edge (∼0.9 S⊕, * brightness of the host star, assuming cloud-free atmospheres: again for low-mass stars). We extend our sample to include 3 planets substantially inward of the nominal habitable zone RTp eq TSM=´´() Scale factor 10-mJ 5 . () 1 boundary because these planets are crucial for testing both the 2 MRp concept of the habitable zone (Bean et al. 2017) and theories of * atmospheric evolution that are relevant for potentially habitable The quantities in Equation (1) are defined as follows: planets (Schaefer et al. 2016; Morley et al. 2017). 1. Rp: the radius of the planet in units of Earth radii, 2. Mp: the mass of the planet in units of Earth masses, 2.3. Emission Sample which, if unknown, should be calculated using the empirical mass–radius relationship of Chen & Kipping Morley et al. (2017) have recently suggested that thermal (2017) as implemented by Louie et al. (2018), emission measurements at long wavelengths (i.e., λ>5 μm) 3.58 with JWST/MIRI could be more insightful than transmission MRRRp =<0.9718p forp 1.23Å , measurements for the warmer terrestrial planets for a given 1.70 MRp =<<1.436p for 1.23 RRp 14.26Å ,() 2 observing time. Therefore, we also estimate the emission spectroscopy secondary eclipse S/N for the Sullivan et al. 3. R*: the radius of the host star in units of solar radii, ’ (2015) planets in the terrestrial bin as a special sample. There is 4. Teq: the planet s equilibrium temperature in Kelvin no MIRI emission spectroscopy analog of the Louie et al. calculated for zero albedo and full day-night heat (2018) paper, so we consider all the terrestrial planets in the redistribution according to Sullivan et al. (2015) catalog and scale the S/N calculations to ⎛ ⎞14 R ⎜⎟1 the expected secondary eclipse depth for a well-studied TTeq = * ,3() * a ⎝ 4 ⎠ example planet (see Section 3.2). We refer to the targets in this group as the “emission” sample. where T* is the host star effective temperature in Kelvin, The choice here to focus on just terrestrial planets for and a is the orbital semimajor axis given in the same units emission spectroscopy is not to say that such observations are as R*, not interesting for larger planets, but rather that a full scale 5. mJ: the apparent magnitude of the host star in the J band, MIRI S/N estimate of secondary eclipses for the Sullivan et al. chosen as a filter that is close to the middle of the NIRISS (2015) catalog is beyond the scope of this paper. We choose to bandpass. focus on the terrestrial planets as there is a particular need to The “scale factor” in Equation (1) is a normalization constant identify additional planets to target with JWSTʼs unique selected to give one-to-one scaling between our analytic capabilities to study small signals at long wavelengths. transmission metric and the more detailed work of Louie Furthermore, larger planets that are good emission spectrosc- et al. (2018) for their 10 hr simulations (with half of that time opy targets will generally also be good targets for transmission occurring during transit). The scale factor also absorbs the unit spectroscopy because both methods have similar dependencies conversion factors so that the parameters can be in natural on planet and host star size and planet temperature, although units. By including this normalizing factor, Equation (1) reports emission spectroscopy has a much steeper dependency on the near-realistic values of the expected S/N for 10 hr observing latter. programs with JWST/NIRISS, modulo our assumptions on atmospheric composition, cloud-free atmospheres, and a fixed mass–radius relation. We determine the scale factor separately 3. Analysis for each planet radius bin using the average of the planets with In this section we write down analytic metrics for the mJ>9 (see Section 4.1 for a discussion of the metric’s expected S/N of transmission and emission spectroscopy applicability to bright stars). The resulting values are given in observations. The transmission spectroscopy metric is applied Table 1. to the statistical and small temperate planet samples, and the Our transmission metric, and its function of selecting the top emission spectroscopy metric is applied to the terrestrial planet atmospheric characterization targets, is similar to that of Zellem emission sample. et al. (2017). Morgan et al. (2018) have also developed a

5 Publications of the Astronomical Society of the Pacific, 130:114401 (14pp), 2018 November Kempton et al.

Table 1 TSM Values and Associated Scale Factors for the Statistical Samplea

Rp<1.5 R⊕ 1.5<Rp<2.75 R⊕ 2.75<Rp<4.0 R⊕ 4.0<Rp<10 R⊕ First quartile (top 25) L 178 146 159 Second quartile (rank 25–50) L 125 124 96 Third quartile (rank 50–75) L 109 95 51 Fourth quartile (rank 75–100) L 92 84 12 Top quintile (N=37) 12 LLL

Scale factor 0.190 1.26 1.28 1.15

Note. a The bold numbers indicate our suggested cutoffs for follow-up efforts. ranking metric, although there are key differences in the The second and third terms in Equation (4) provide the implementation (e.g., the use of an explicit mass–radius appropriate scaling of the secondary eclipse depth at a relation, and no order unity correction for transit duration). wavelength of 7.5 μm, which is the center of the MIRI LRS By not including a factor of the mean molecular weight, μ,in bandpass (5–10 μm, Kendrew et al. 2015; Rieke et al. 2015). Equation (1), we implicitly assume that all planets in a given The final term scales the S/N by the K-band flux of the planet’s size bin have the same atmospheric composition. This is the host star. We chose the K-band magnitude for the ESM because same assumption made by Louie et al. (2018), who chose it is the longest wavelength photometric magnitude that is values of μ=18 (in units of the proton mass) for planets with given in the Sullivan et al. (2015) catalog. 6 Rp<1.5 R⊕, and μ=2.3 for planets with Rp>1.5 R⊕. The The factor of 4.29×10 in front of Equation (4) scales the former is consistent with a water (steam) atmosphere and the ESM to yield a S/N of 7.5 in the MIRI LRS bandpass for a latter with a solar composition, H2-dominated atmosphere. Any single secondary eclipse of the reference planet GJ 1132b intrinsic spread in the atmospheric composition (expected (Berta-Thompson et al. 2015), based on detailed modeling of primarily in the smaller planet size bins) will translate linearly its atmosphere described below. We chose GJ 1132b as the into the S/N realized in an actual JWST observing campaign. reference because Morley et al. (2017) have shown that it is the For the calculations of the small temperate sample we rescale best known small exoplanet for thermal emission measure- the Louie et al. (2018) S/N and our metric for the ments with JWST. We also confirm that many of the small < < / 1.5 Rp 2.0 R⊕ planets by a factor of 2.3 18 to put these TESS planets that are likely good targets are similar to this planets on the same basis as the smaller planets, in terms of planet. Throughout this analysis we assume the Dittmann et al. their mean molecular weights. That is, we assume for the (2017) values for the properties of GJ 1132b and its host star purpose of investigating that the planets (Rp/R*=0.0455, a/R*=16.54, T*=3270 K). in question all have dense, secondary atmospheres. The emission metric assumes that both the planet and the star emit as blackbodies. For stars this tends to be a reasonable − 3.2. Emission Metric assumption at mid-IR wavelengths, although continuum H opacity combined with line blanketing at shorter wavelengths For the planets in the terrestrial bin we also compute an results in infrared brightness temperatures that differ from the emission spectroscopy metric (ESM) that is proportional to the effective temperature. For example, for the benchmark planet expected S/NofaJWST secondary eclipse detection at mid-IR GJ 1132b we find that models predict that its host star’s wavelengths. The metric (which is also similar to the one from brightness temperature is 90% of its effective temperature in Zellem et al. 2018) is the MIRI LRS bandpass (i.e., 2922 versus 3270 K). However, ⎛ ⎞2 since this factor is different for varying stellar types we elect BT7.5() day Rp ESM=´´ 4.29 106 ´⎜ ⎟ ´ 10-mK 5 .() 4 not to apply a correction to the ESM beyond the normalization ⎝ ⎠ BT7.5()**R factor that is already applied in Equation (4). We additionally performed tests that show that the relative scaling of planets The new quantities in Equation (4) are defined as follows: according to the ESM in Equation (4) is not sensitive to 10%

1. B7.5: Planck’s function evaluated for a given temperature changes in the stellar temperatures. at a representative wavelength of 7.5 μm, The assumption of blackbody emission is more problematic 2. Tday: the planet’s dayside temperature in Kelvin, which for the planets because their emergent mid-IR spectra are we calculate as 1.10×Teq, and strongly sculpted by molecular absorption and the emitted 3. mK: the apparent magnitude of the host star in the K band. flux can vary by an order of magnitude or more between

6 Publications of the Astronomical Society of the Pacific, 130:114401 (14pp), 2018 November Kempton et al. spectral bands and low opacity windows that probe the deep secondary eclipse depth of 75 ppm for GJ 1132b binned over atmosphere. However, the single temperature blackbody the MIRI LRS bandpass. This estimate is consistent with the assumption is reasonable for a S/N metric that aims to convey predictions of Morley et al. (2017) modulo the different the relative broadband observability of planets. We verify that assumptions of albedo and redistribution. our blackbody assumption gives an intermediate prediction of Finally, we estimated the noise on a single broadband secondary the emission signal for different plausible atmospheric eclipse measurement of GJ 1132b with MIRI LRS using the compositions in a specific example below. PandExo simulation tool (Batalha et al. 2017b). We determined a The ESM is a stronger function of the assumed planetary photon-limited error of 10 ppm from this calculation, which yields temperature than the stellar temperature because observations a7.5σ detection of GJ 1132b accordingtothemodelsdescribed at 7.5 μm are nearer to the planets’ peak of blackbody above. Our predicted significance is substantially less than that emission. Observations at secondary eclipse probe exoplanets’ givenbyMorleyetal.(2017) forsimilarmodelsduetoanerrorin daysides. We therefore apply a theoretically derived correction those authors’ calculations (L. Kreidberg & C. Morley 2018, factor (1.10, see below) to the equilibrium temperature for personal communication).WealsonotethattheJWST throughput estimating the dayside temperature needed to predict the numbers in PandExo have evolved over the last year due to the secondary eclipse depth. incorporation of the latest instrument testing data. Our PandExo We performed theoretical calculations for the atmosphere of simulation is from February 2018; future simulations may find GJ 1132b using 3D GCMs and 1D radiative-convective forward different results if the assumptions in PandExo change. models to investigate the S/N scaling factor needed for the ESM (D. D. B. Koll et al. 2018, in preparation). First, we used 3D 4. Results GCMs to investigate the energy transport for GJ 1132b and similar synchronously rotating planets that TESS will find. The 4.1. Statistical Sample GCM is the same as described in Koll & Abbot (2015, 2016), In Figure 3 we plot the 10 hr JWST/NIRISS S/N from Louie and solves the equations of atmospheric motion coupled to gray et al. (2018) versus the TSM from Equation (1) for each planet radiative transfer. We assume the atmosphere is transparent size bin in the statistical sample. As can be seen, the analytic to stellar radiation and that its infrared opacity is comparable metric tracks the S/N calculated by Louie et al. (2018) with ( to representative values for the Solar System Robinson & little scatter for targets with mJ>9. Simulated planets with Catling 2014), and we investigate the atmosphere’sheat particularly bright host stars exhibit a different slope in the redistribution as a function of surface pressure. From the GCM relationship due to differences in the observational duty cycle calculations we find that a 1 bar atmosphere will have moderate with JWST that our metric does not capture. In the Louie et al. heat redistribution, consistent with a conventional redistribution (2018) simulations, stars with mJ∼8.1 (depending on the factor of f=0.53 (where f=1/4 is full planet redistribution stellar type, for more details see the University of Montreal and f=2/3 is instant re-radiation). JWST website52 and Beichman et al. 2014) require the use of ’ We also calculated 1D forward models to estimate GJ 1132b s the bright star readout mode, and stars with mJ<9 start to thermal emission signal for different compositions as a check of have substantially lower duty cycles due to the limited number the ESM assumptions and to provide an absolute S/N benchmark. of reads possible before a reset of the detector is needed. These include “double-gray” calculations of the planet’stemper- We did not attempt to correct for the mismatch between the ature-pressure profile combined with a wavelength-dependent Louie et al. (2018) results and the TSM for bright stars because solution of the radiative transfer equation (without scattering) to the TSM is intended as a general metric for the ranking of predict the planet’s emission spectrum, as described in Miller-Ricci transmission spectroscopy targets for infrared observations. A et al. (2009). The calculations were done for a redistribution factor correction factor of the square root of the duty cycle can be of 0.53 and a surface pressure of 1 bar for consistency with the applied to account for duty-cycle reductions for e.g bright stars GCM results. We also adopted an Earth-like albedo of 0.3 absent with JWST/NIRISS. Furthermore, systems with bright host any empirical constraints on the characteristics of terrestrial stars have benefits that balance against their non-ideal nature exoplanet atmospheres. This combination of albedo and redis- for JWST observations. For example, bright stars make RV tribution results in a dayside temperature that is 10% higher than mass measurements easier. Also, dedicated missions like the full redistribution equilibrium temperature calculated from ARIEL have much smaller apertures than JWST and will ( ) = × Equation 3 . We therefore adopt a scaling of Tday 1.10 Teq therefore only suffer reduced duty cycles for extremely bright for our ESM calculations. stars. Bright stars are also typically preferred for ground-based We considered three atmospheric compositions in our 1D high resolution observations due to the higher background in modeling: H2-rich solar composition gas; pure H2O steam; and these data, the possibility of using adaptive optics systems to a Venus-like composition of 96.5% CO2, 3.5% N2, plus trace reduce slit losses, and the capability of these facilities and amounts of H2O and CO. From averaging the results for the three different types of atmospheres, we estimate a typical 52 http://jwst.astro.umontreal.ca/?page_id=51

7 Publications of the Astronomical Society of the Pacific, 130:114401 (14pp), 2018 November Kempton et al.

Figure 3. NIRISS S/N from Louie et al. (2018) vs. the transmission spectroscopy metric from Equation (1), using the scale factors from Table 1. The points are simulated TESS planets (black: mJ>9, red: mJ<9). The dashed line plots a one-to-one relationship. The brighter stars likely deviate from the one-to-one relationship because they have lower duty cycle for JWST observations, which our analytic metric does not capture. The TSM values for known benchmark planets are indicated by the x-axis position of the vertical blue lines, assuming these planets have the same atmospheric composition as the rest of the sample. (A color version of this figure is available in the online journal.) instruments to observe very bright stars without duty cycle TSM between the second and third quartiles for the sub-Jovian penalties. Furthermore, higher efficiency read modes for JWST bin belies their relative scarcity at the orbital periods TESS is observations of bright stars are currently being investigated sensitive to. For this reason, we suggest selecting the top 50 sub- (Batalha et al. 2018). Jovian planets as the high-priority atmospheric characterization In Table 1 we give the cutoff values of the TSM for the top targets within that bin. quintile of terrestrial planets and the top 100 planets (sub-divided The values in Table 1 can be used to prioritize observations into 25-planet groupings) for the three largest size bins in the aimed at confirming and measuring the masses of potential statistical sample. Both the large and small sub-Neptune bins atmospheric characterization targets. For example, planets have a plethora of targets that would yield high S/N atmospheric found early in the TESS mission with smaller metric values characterization with JWST. However, the dramatic fall off in the than those reported in boldface in Table 1 for the relevant bins

8 Publications of the Astronomical Society of the Pacific, 130:114401 (14pp), 2018 November Kempton et al.

Figure 4. NIRISS S/N from Louie et al. (2018) vs.the transmission spectroscopy metric from Equation (1) for the small temperates, following the same labeling conventions as Figure 3. The left panel assumes the Chen & Kipping (2017) empirical mass–radius relationship for the simulated planets and the currently measured masses for the known planets. The right panel assumes that all the planets have Earth-like composition with their masses predicted by the formula given by Zeng et al. (2016) with a core mass fraction of 0.3. (A color version of this figure is available in the online journal.)

Table 2 Top 10 Habitable Zone Planets—Empirical Planet Masses

a b Name TSM Mp Rp M* R* RV SmV mJ −1 (M⊕)(R⊕)(Me)(Re)(ms )(S⊕) TESS-Sim 204 27.9 0.39 0.77 0.24 0.24 0.27 1.27 11.50 7.91 TESS-Sim 1296 26.8 4.35 1.92 0.17 0.17 4.04 1.00 13.90 10.00 TESS-Sim 1804 26.5 4.42 1.94 0.16 0.16 3.64 0.45 14.30 9.78 TESS-Sim 1308 23.2 2.83 1.49 0.25 0.25 1.80 1.15 11.60 7.97 TESS-Sim 922 21.6 2.53 1.39 0.12 0.12 4.16 1.17 16.20 11.26 TESS-Sim 405 19.4 3.11 1.58 0.38 0.38 1.33 1.61 10.10 6.85 TESS-Sim 105 17.9 3.48 1.68 0.14 0.14 4.15 1.12 15.30 11.27 TESS-Sim 48 17.3 4.64 1.99 0.16 0.16 4.38 0.64 15.00 11.10 TESS-Sim 1244 16.8 3.99 1.82 0.16 0.16 4.45 1.79 15.30 11.34 TESS-Sim 991 15.8 3.67 1.74 0.16 0.16 2.97 0.39 14.40 10.53

LHS 1140b 9.8 6.98 1.73 0.15 0.19 5.42 0.39 14.15 9.61 TRAPPIST-1f 23.3 0.69 1.05 0.09 0.12 1.05 0.36 18.80 11.35

Notes. a Planet names from the simulated Sullivan et al. (2015) TESS catalog. b Scale factor=0.167, calculated for the small temperate sample.

could be confidently set aside in favor of better planets that will 4.2. Small Temperate Sample be found as the mission progresses. On the other hand, planets with metric values near the top should be seen as high priorities The results for the planets in and near the habitable zone are for follow-up efforts with the expectation that few other planets shown in the left panel of Figure 4. The analytic TSM performs that are as good atmospheric characterization targets will be very well for these simulated planets compared to the Louie found later. et al. (2018) results, again with the exception of the small

9 Publications of the Astronomical Society of the Pacific, 130:114401 (14pp), 2018 November Kempton et al.

Table 3 Top 10 Habitable Zone Planets—Rocky Planet Masses

a b Name TSM Mp Rp M* R* RV SmV mJ −1 (M⊕)(R⊕)(Me)(Re)(ms )(S⊕) TESS-Sim 204 28.2 0.38 0.77 0.24 0.24 0.27 1.27 11.50 7.91 TESS-Sim 922 16.2 3.36 1.39 0.12 0.12 5.52 1.17 16.20 11.26 TESS-Sim 1308 15.1 4.34 1.49 0.25 0.25 2.76 1.15 11.60 7.97 TESS-Sim 405 11.2 5.40 1.58 0.38 0.38 2.31 1.61 10.10 6.85 TESS-Sim 1878 11.0 1.32 1.08 0.14 0.14 1.44 0.78 15.40 11.37 TESS-Sim 1296 10.5 11.10 1.92 0.17 0.17 10.30 1.00 13.90 10.00 TESS-Sim 1804 10.2 11.53 1.94 0.16 0.16 9.49 0.45 14.30 9.78 TESS-Sim 45 9.9 1.19 1.05 0.16 0.16 1.01 0.62 14.80 10.91 TESS-Sim 1292 9.6 4.68 1.52 0.25 0.25 3.14 1.35 12.70 9.06 TESS-Sim 105 9.2 6.77 1.68 0.14 0.14 8.07 1.12 15.30 11.27

LHS 1140b 9.15 7.50 1.73 0.15 0.19 5.83 0.39 14.15 9.61 TRAPPIST-1f 13.7 1.05 1.05 0.09 0.12 1.56 0.36 18.80 11.35

Notes. a Planet names from the simulated Sullivan et al. (2015) TESS catalog. b Scale factor=0.167, calculated for the small temperate sample.

number of planets orbiting stars brighter than mJ=9. The very small stars (note that all the host stars in Tables 2 and 3 are TSM values and planet parameters for the top 10 simulated late M dwarfs). The detection of systems like LHS 1140 and planets are given in Table 2, benchmarked against the known TRAPPIST-1 with ground-based instruments suggests that planets LHS 1140b (Dittmann et al. 2017; Ment et al. 2018) perhaps the assumed occurrence rates in this regime are and TRAPPIST-1f (Gillon et al. 2017; Grimm et al. 2018). For underestimated and that TESS will find more of these nearby the simulated TESS planets we estimate the stellar masses systems. Also, further photometric monitoring to search for assuming a one-to-one relationship between stellar radius and additional transiting planets in TESS-discovered systems may mass, as indicated by Boyajian et al. (2012), and compute the boost the yield of potentially habitable planets (Ballard 2018). radial velocity signal consistent with our planet masses. Based on the values reported in Tables 2 and 3 we therefore ∼ ( The Chen & Kipping (2017) masses assumed for this study propose that a TSM value of 10 assuming either the – ) imply a significant volatile component for the larger planets in empirical mass radius relation or rocky composition for all the small temperate sample, and thus the observability of these represents a good starting threshold for evaluating the atmo- fi planets in the context of habitability may be overestimated spheric observability of potentially habitable planets identi ed compared to that of the smaller planets. Therefore, we repeat with TESS. Small planet candidates receiving Earth-like insolation and having TSM values substantially larger than our analysis for this sample with a re-scaling of the Louie et al. this cutoff should be high priority targets for follow-up efforts (2018) results and the TSM assuming all the planets have in the context of future atmospheric characterization. Planets Earth-like composition. Here we estimate the planet masses discovered early in the mission that are near or below this from their radii using the formula given by Zeng et al. (2016) threshold value should only become high priority targets for with a core mass fraction of 0.3 (i.e., MR= 0.993 3.7, where p p follow-up observations for reasons other than atmospheric ) Mp and Rp are in Earth units . We also recompute the metric for observability (e.g., exploring the mass–radius relationship for the planets LHS 1140b and TRAPPIST-1f with the same temperate planets). Ultimately, the TSM cutoff values for assumption of Earth-like composition. The results of this prioritization for each of the samples should be re-evaluated calculation are shown in the right panel of Figure 4 and the once we get a handle on the actual TESS yield, say after the first parameters for the recomputed top 10 simulated planets are year of the mission. given in Table 3. As previously pointed out by Louie et al. (2018), although TESS will indeed find small planets in the habitable zones of 4.3. Emission Sample their host stars, the current expectation is that only a few of Table 4 identifies the set of 20 targets that have emission these planets will be as good or better targets for atmospheric spectroscopy metric values larger than that of GJ 1132b characterization than the currently known planets. However, as (ESM=7.5). Our benchmark planet GJ 1132b is currently the also pointed out by Louie et al. (2018), this prediction hinges best of the known small planets for secondary eclipse measure- critically on the assumed frequency of small planets around ments at mid-IR wavelengths according to Morley et al. (2017).

10 Publications of the Astronomical Society of the Pacific, 130:114401 (14pp), 2018 November Kempton et al.

Table 4 Top Emission Spectroscopy Targets

a b Name ESM TSM Mp Rp Tp Porb M* R* T* RV mV mK −1 (M⊕)(R⊕)(K)(days)(Me)(Re)(K)(ms ) TESS-Sim 284 112.5 107.4 2.63 1.43 980 0.6 0.23 0.23 2560 5.36 13.10 6.80 TESS-Sim 1763 29.8 49.8 2.46 1.37 1749 1.1 0.81 0.81 5190 1.75 6.47 4.47 TESS-Sim 1476 26.8 77.9 0.28 1.38 661 2.3 0.29 0.29 3450 2.70 9.70 5.54 TESS-Sim 21 23.6 35.9 2.53 1.49 1672 1.0 0.70 0.70 5030 2.27 8.00 5.85 TESS-Sim 1855 14.7 23.6 1.31 1.35 990 1.1 0.42 0.42 3640 2.61 11.30 7.36 TESS-Sim 1957 11.7 25.2 2.44 1.14 1815 1.0 0.81 0.81 5160 1.14 7.84 5.82 TESS-Sim 1745 11.5 23.6 2.01 1.09 994 0.5 0.24 0.24 3340 2.69 14.00 9.59 TESS-Sim 1421 10.8 15.8 2.26 1.40 950 0.6 0.27 0.27 3360 4.55 14.50 10.10 TESS-Sim 858 10.5 12.4 1.23 1.48 1210 0.6 0.41 0.41 3590 3.87 13.60 9.53 TESS-Sim 1255 9.9 22.4 0.39 1.22 697 0.6 0.14 0.14 2870 5.70 16.40 11.30 TESS-Sim 675 9.6 14.1 1.18 1.47 905 0.6 0.21 0.21 3270 6.12 16.20 11.60 TESS-Sim 1926 9.0 13.1 1.93 1.38 1028 0.8 0.37 0.37 3520 3.31 13.50 9.34 TESS-Sim 1340 8.4 12.0 2.39 1.44 1084 1.1 0.48 0.48 3760 2.71 12.40 8.60 TESS-Sim 289 8.2 9.9 2.83 1.49 1221 0.7 0.47 0.47 3740 3.37 13.30 9.48 TESS-Sim 90 8.2 13.4 1.42 1.34 1007 1.4 0.48 0.48 3770 2.23 11.80 8.00 TESS-Sim 419 8.2 15.9 2.58 1.09 1111 0.6 0.38 0.38 3530 1.90 12.90 8.81 TESS-Sim 1780 8.2 11.6 2.56 1.26 1072 0.7 0.36 0.36 3500 3.08 13.60 9.48 TESS-Sim 884 7.6 9.1 2.36 1.32 1298 0.5 0.43 0.43 3640 3.24 13.70 9.78 TESS-Sim 1160 7.6 9.8 1.10 1.26 1265 0.5 0.43 0.43 3630 2.95 13.50 9.52 TESS-Sim 1962 7.5 8.8 2.83 1.32 1342 0.5 0.46 0.46 3690 3.09 13.60 9.66

Notes. a Planet names from the simulated Sullivan et al. (2015) TESS catalog. b TSM calculated with a scale factor of 0.190.

Even so, this planet is not guaranteed to be a straightforward cycle for very bright stars, as we did for the TSM (although target for emission spectroscopy with JWST. With a predicted note that the MIRI detector is more efficient than the NIRISS S/N in the MIRI LRS bandpass-integrated secondary eclipse of detector, Batalha et al. 2017b). And there is the issue that 7.5, it will take at least two eclipses and photon-limited these bright systems require higher photon-limited precision performance of the instrument to build up a “white light” S/N that might not be obtainable in the face of instrument of greater than 10. This should be seen as the minimum systematics. On the other hand, MIRI becomes background requirement for secondary eclipse spectroscopy, which further limited for stars fainter than mJ=10 (Batalha et al. 2018). reduces the S/N by dividing the observation into smaller Therefore, the S/N of real observations of some of the fainter wavelength intervals. Furthermore, the single eclipse S/Nfor targets in Table 4 would be lower than what is predicted from GJ 1132b of 7.5 means that the secondary eclipse will easily be the ESM. recoverable from a single epoch of observations—an important The other targets in Table 4 have emission spectroscopy S/N consideration for planets whose secondary eclipse timing is not values that are marginally better than that of GJ 1132b. As with well constrained due to uncertainties on orbital eccentricities or the small temperate sample, the emission sample primarily fi ephemerides. We therefore suggest that GJ 1132b’s ESM of 7.5 identi es planets from within a region of parameter space ( should be selected as the cutoff value in identifying the top where planet occurrence rates have high uncertainties i.e., very ) emission spectroscopy small planets for JWST. small host stars and small planets on ultra-short period orbits . As expected, the planets in Table 4 are distinguished by We therefore caution that it is reasonable to expect the actual emission sample from TESS will vary in size from our high Teq,smallR*,and/or bright host stars. All 20 planets have higher equilibrium temperatures than GJ 1132b, affirm- prediction of 20 planets. ing that this is likely to remain one of the best (if not the best) 5. Discussion and Conclusion planets for thermal emission measurements with Teq<600 K in perpetuity. Of the 20 planets identified in Table 4,thetop6 In summary, we have suggested simple, analytic metrics for would be very challenging targets for JWST due to the determining which transiting exoplanets are the best targets for brightnesses of their host stars. Furthermore, the ESM likely atmospheric characterization, with a focus on JWST capabil- overestimates the S/N of real observations for these systems ities. Applying these metrics specifically to the expected TESS because we neglect to consider the impact of the reduced duty yield, we have determined appropriate cutoff values to identify

11 Publications of the Astronomical Society of the Pacific, 130:114401 (14pp), 2018 November Kempton et al.

ground-based observations to the host star brightness. The metric calculations can also be applied to existing exoplanet candidates such as those from the K2 mission to identify high- quality atmospheric characterization targets that will rival those that are expected to be discovered by TESS. We have also repeated our analysis using the simulated TESS catalog of Barclay et al. (2018), to assess the impact of using an independent realization of the mission outcome. We find that applying the same set of threshold TSM criteria from Figure 5 also results in a statistical sample of 250–300 planets, although the radius distribution of those objects is somewhat altered. The Barclay et al. (2018) catalog returns more sub-Jovians and fewer sub-Neptunes (from both size bins from the latter category). The larger number of sub-Jovians likely results from the inclusion of the TESS full frame images in generating the Barclay et al. (2018) catalog, whereas the main Sullivan et al. (2015) catalog only accounted for the 2-minute cadence targets. Somewhat surprisingly, despite the known overestimation of the number of terrestrial planets in Sullivan et al. (2015), the Barclay et al. (2018) catalog actually produces several more terrestrials above our suggested TSM threshold value of 10. That is to say that while Barclay et al. (2018) predict fewer overall terrestrial planets, they produce higher quality targets Figure 5. Summary of the properties and threshold metric values for the various atmospheric characterization samples described in this paper. with respect to transmission spectroscopy observations. (A color version of this figure is available in the online journal.) In addition to the selection of top atmospheric characteriza- tion targets using the TSM and ESM threshold values, additional factors may play into the decision to further prioritize or de-prioritize individual targets. Refinement of the planets that should be advanced expeditiously for RV follow- sample is also advised based on factors such as expected RV up and subsequent atmospheric investigations. For the purpose amplitude, stellar activity level, high false positive likelihood of selecting easy-to-remember round numbers for the threshold (e.g., near-grazing transits), JWST observability (i.e., prioritizing transmission spectroscopy metric, based on the values in targets that lie within the JWST continuous viewing zone),and Table 1, we recommend that planets with TSM >10 for the precision to which ephemerides and other system parameters Rp<1.5 R⊕ and TSM >90 for 1.5<Rp<10 R⊕ be selected are known. Furthermore, while our aim has been to develop a as high-quality atmospheric characterization targets among the truly statistical sample of exoplanet atmospheres, we acknowl- TESS planetary candidates. We also recommend a threshold of edge the biases that will ultimately remain in the selected targets. TSM=10 for putative habitable zone planets. For emission For example, the terrestrial and small sub-Neptune samples are spectroscopy of terrestrial planets, we recommend a threshold heavily dominated by planets orbiting M stars because of their of ESM=7.5. Applying these cuts should result in ∼300 new relatively larger transit depths, whereas the sub-Jovian sample is ideal targets for transmission spectroscopy investigations from weighted toward Sun-like hosts owing to the intrinsic scarcity of the TESS mission. such planets around smaller M dwarfs. Targets that buck these We review the various atmospheric characterization samples trends in host star type should therefore also be prioritized more in Figure 5, along with their metric selection criteria. We note highly. from Tables 2–4 that both the small temperate and emission The similar metric cutoff values for the sub-Neptune and samples are almost fully contained within the statistical sample. sub-Jovian planets reflects the fact that the Louie et al. (2018) Therefore, the selection criteria for the statistical sample should simulation predicts similar S/N across a wide range of be seen as the primary mode for identifying appropriate JWST planetary sizes due to their adoption of the same atmospheric atmospheric characterization targets from the TESS returns. mean molecular weight and not accounting for the potential Furthermore, the same TSM and ESM calculations can be used impact of clouds, which admittedly are difficult to impossible to identify high priority targets for exoplanet atmosphere to accurately predict a priori. Crossfield & Kreidberg (2017) studies with other facilities such as ARIEL and ground-based have pointed out that smaller and cooler exoplanets empirically ELTs, although in the latter case the threshold criteria may have smaller features in their transmission spectra relative to need to be revised to account for the stronger sensitivity of expectations. This suggests that such planets have higher mean

12 Publications of the Astronomical Society of the Pacific, 130:114401 (14pp), 2018 November Kempton et al.

Figure 6. RV semi-amplitude vs.NIRISS S/N for the planets in the simulated Sullivan et al. (2015) TESS catalog and the Louie et al. (2018) 10 hr S/N predictions. Filled triangles denote the planets included in our transmission statistical sample using the threshold criteria from Table 1, upside down triangles indicate the planets included in our emission sample, and small x’s denote targets that are disfavored for atmospheric characterization based on low expected S/N. (Note that the S/N values were calculated for a high-μ water-rich atmosphere for planets with Rp<1.5 R⊕ and a low-μ hydrogen-rich atmosphere for Rp>1.5 R⊕.) (A color version of this figure is available in the online journal.)

molecular weight atmospheres and/or an increased prevalence challenging for RV mass measurements owing to the small of high altitude aerosols. Both of these phenomena would signals and relative faintness of the host stars. reduce the expected S/N from the nominal calculations of Despite the obvious challenges, we propose that a large-scale Louie et al. (2018). So while the TSM is useful for prioritizing effort to confirm and precisely measure the masses of hundreds targets, the allocation of telescope time for atmospheric of planets detected by TESS is well justified. A key strength of characterization will need to be carefully matched to the exoplanet studies is the chance to perform statistical investiga- specific planets to be observed and the scientific objectives of tions that are not possible with the limited sample of planets in the program. the Solar System. We are of the opinion that the study of The predicted RV semi-amplitudes of the simulated TESS exoplanet atmospheres is no different in this regard than studies planets in the statistical sample are shown in Figure 6 (note that of planetary frequency (Bean et al. 2017). Furthermore, the the stellar masses and resulting radial velocity signals are study of a large sample of sub-Neptune exoplanet atmospheres calculated as described in Section 4.2). The bulk of the RV is especially important because these planets do not exist in our signals for the best targets identified in this study range Solar System, and therefore no well-studied benchmark objects between 1–10 m s−1, which is within reach for many instru- exist. We also argue that a larger sample of sub-Neptune ments that already exist or are currently under construction. planets is needed than for giant planets due to the higher degree However, it is worth acknowledging that currently only ∼250 of diversity expected for these atmospheres in terms of their known planets have both masses and radii measured to 10% or bulk compositions, which is a natural outcome of our proposed better precision. Therefore, the goal laid out here of RV follow TESS follow-up strategy. up for hundreds of new TESS planets, most of which have Preliminary results suggest that measuring masses for ∼300 Rp<4 R⊕, may seem overly ambitious given the resource- of the best TESS planets for atmospheric characterization intensive nature of RV observations. In addition, each of the would require approximately 400 nights of observing time small temperate planets listed in Tables 2 and 3 will be very (Cloutier et al. 2018). Ultimately, we are optimistic that the

13 Publications of the Astronomical Society of the Pacific, 130:114401 (14pp), 2018 November Kempton et al. large number of high-precision RV instruments expected to Batalha, N. E., Lewis, N. K., Line, M. R., Valenti, J., & Stevenson, K. 2018, come online within the next few years (Fischer et al. 2016; ApJL, 856, L34 ) Batalha, N. E., & Line, M. R. 2017, AJ, 153, 151 Wright & Robertson 2017 will bring the goal of dramatically Batalha, N. E., Mandell, A., Pontoppidan, K., et al. 2017b, PASP, 129, 064501 expanding the sample of atmospheric characterization targets Bean, J. & FINESSE Science Team 2017, AAS Meeting, 229, 301.08 within reach. Bean, J. L., Abbot, D. S., & Kempton, E. M.-R. 2017, ApJL, 841, L24 Beichman, C., Benneke, B., Knutson, H., et al. 2014, PASP, 126, 1134 Berta-Thompson, Z. K., Irwin, J., Charbonneau, D., et al. 2015, Natur, The work of E.M.-R.K. was supported by the National 527, 204 Science Foundation under grant No. 1654295 and by the Bouma, L. G., Winn, J. N., Kosiarek, J., & McCullough, P. R. 2017, arXiv:1705.08891 Research Corporation for Science Advancement through their Boyajian, T. S., von Braun, K., van Belle, G., et al. 2012, ApJ, 757, 112 Cottrell Scholar program. J.L.B. acknowledges support from Chen, J., & Kipping, D. 2017, ApJ, 834, 17 the David and Lucile Packard Foundation and NASA through Cloutier, R., Doyon, R., Bouchy, F., & Hébrard, G. 2018, AJ, 156, 82 Cowan, N. B., Greene, T., Angerhausen, D., et al. 2015, PASP, 127, 311 STScI grants GO-14792 and 14793. D.R.L. acknowledges Crossfield, I. J. M., & Kreidberg, L. 2017, AJ, 154, 261 support from NASA Headquarters under the NASA Earth Dittmann, J. A., Irwin, J. M., Charbonneau, D., et al. 2017, Natur, 544, 333 and Space Science Fellowship (NESSF) Program—Grant Dressing, C. D., & Charbonneau, D. 2015, ApJ, 807, 45 Fischer, D. A., Anglada-Escude, G., Arriagada, P., et al. 2016, PASP, 128, NNX16AP52H. D. Deming acknowledges support from the 066001 TESS mission. Part of the research was carried out at the Jet Fortney, J. J., Mordasini, C., Nettelmann, N., et al. 2013, ApJ, 775, 80 Propulsion Laboratory, California Institute of Technology, Fressin, F., Torres, G., Charbonneau, D., et al. 2013, ApJ, 766, 81 Fulton, B. J., Petigura, E. A., Howard, A. W., et al. 2017, AJ, 154, 109 under contract with the National Aeronautics and Space Gillon, M., Triaud, A. H. M. J., Demory, B.-O., et al. 2017, Natur, 542, Administration. D.D.B.K. was supported by a James McDon- 456 nell Foundation postdoctoral fellowship. J.K.B. is supported by Girardi, L., Groenewegen, M. A. T., Hatziminaoglou, E., & da Costa, L. 2005, A&A, 436, 895 a Royal Astronomical Society Research Fellowship. T.B. and Grimm, S. L., Demory, B.-O., Gillon, M., et al. 2018, A&A, 613, A68 E.V.Q. are grateful for support from GSFC Sellers Exoplanet Howe, A. R., Burrows, A., & Deming, D. 2017, ApJ, 835, 96 Environments Collaboration (SEEC). D.C. acknowledges Huang, C. X., Shporer, A., Dragomir, D., et al. 2018, arXiv:1807.11129 Kane, S. R., Mahadevan, S., von Braun, K., Laughlin, G., & Ciardi, D. R. support from the John Templeton Foundation. The opinions 2009, PASP, 121, 1386 expressed here are those of the authors and do not necessarily Kendrew, S., Scheithauer, S., Bouchet, P., et al. 2015, PASP, 127, 623 reflect the views of John Templeton Foundation. D. Dragomir Koll, D. D. B., & Abbot, D. S. 2015, ApJ, 802, 21 Koll, D. D. B., & Abbot, D. S. 2016, ApJ, 825, 99 acknowledges support provided by NASA through Hubble Kopparapu, R. K., Ramirez, R., Kasting, J. F., et al. 2013, ApJ, 765, 131 Fellowship grant HST-HF2-51372.001-A awarded by the Louie, D. R., Deming, D., Albert, L., et al. 2018, PASP, 130, 044401 Space Telescope Science Institute, which is operated by the Ment, K., Dittmann, J. A., Astudillo-Defru, N., et al. 2018, arXiv:1808.00485 Miller-Ricci, E., Seager, S., & Sasselov, D. 2009, ApJ, 690, 1056 Association of Universities for Research in Astronomy, Inc., Mordasini, C., van Boekel, R., Mollière, P., Henning, T., & Benneke, B. 2016, for NASA, under contract NAS5-26555. N.N. was partly ApJ, 832, 41 supported by JSPS KAKENHI grant Number JP18H01265 Morgan, J., Kerins, E., Awiphan, S., et al. 2018, arXiv:1802.05645 Morley, C. V., Kreidberg, L., Rustamkulov, Z., Robinson, T., & Fortney, J. J. and JST PRESTO grant Number JPMJPR1775. E.L.S. 2017, ApJ, 850, 121 acknowledges funding from NASA Habitable Worlds grant Ricker, G. R., Winn, J. N., Vanderspek, R., et al. 2015, JATIS, 1, 014003 NNX16AB62G. C.v.E. acknowledges funding for the Stellar Rieke, G. H., Wright, G. S., Böker, T., et al. 2015, PASP, 127, 584 Robinson, T. D., & Catling, D. C. 2014, Nat. Geosci., 7, 12 Astrophysics Centre, provided by The Danish National Rodler, F., & López-Morales, M. 2014, ApJ, 781, 54 Research Foundation (Grant agreement no.: DNRF106). Schaefer, L., Wordsworth, R. D., Berta-Thompson, Z., & Sasselov, D. 2016, ApJ, 829, 63 Snellen, I. A. G., de Kok, R. J., le Poole, R., Brogi, M., & Birkby, J. 2013, ApJ, ORCID iDs 764, 182 // / Stassun, K. G., Corsaro, E., Pepper, J. A., & Gaudi, B. S. 2018, AJ, 155, 22 Eliza M.-R. Kempton, https: orcid.org 0000-0002- Sullivan, P. W., Winn, J. N., Berta-Thompson, Z. K., et al. 2015, ApJ, 809, 1337-9051 77 Jacob L. Bean https://orcid.org/0000-0003-4733-6532 Thompson, S. E., Coughlin, J. L., Hoffman, K., et al. 2018, ApJS, 235, 38 Tinetti, G., Drossart, P., Eccleston, P., et al. 2016, Proc. SPIE, 9904, 99041X Torres, G., Fressin, F., Batalha, N. M., et al. 2011, ApJ, 727, 24 References Van Eylen, V., Agentoft, C., Lundkvist, M. S., et al. 2018, MNRAS, 479, 4786 Wright, J. T., & Robertson, P. 2017, RNAAS, 1, 51 Ballard, S. 2018, arXiv:1801.04949 Zellem, R. T., Fortney, J. J., Swain, M. R., et al. 2018, arXiv:1803.07163 Barclay, T., Pepper, J., & Quintana, E. V. 2018, arXiv:1804.05050 Zellem, R. T., Swain, M. R., Roudier, G., et al. 2017, ApJ, 844, 27 Batalha, N. E., Kempton, E. M.-R., & Mbarek, R. 2017a, ApJL, 836, L5 Zeng, L., Sasselov, D. D., & Jacobsen, S. B. 2016, ApJ, 819, 127

14